I was working with a certain XML where there were no redundancies

<person>
  <eye>
    <eye_info>
       <eye_color>
       blue
       </eye_color>
    </eye_info>
  </eye>
  <hair>
    <hair_info>
       <hair_color>
       blue
       </hair_color>
    </hair_info>
  </hair>
</person>

As you can see, the sub-tag eye-color makes reference to eye in it's name, so there was no need to avoid redundancies, I could get the eye color in a single line after loading the XML into a dataset:

dataset.ReadXml(path);
value = dataset.Tables("eye_info").Rows(0)("eye_color");

I do realise it's not the smartest way of doing so, and this situation I'm having now wasn't unforeseen.

Now, let's say I have to read xml's that are in this format:

<person>
  <eye>
    <info>
       <color>
       blue
       </color>
    </info>
  </eye>
  <hair>
    <info>
       <color>
       blue
       </color>
    </info>
  </hair>
</person>

So If I try to call it like this:

dataset.ReadXml(path);
value = dataset.Tables("info").Rows(0)("color");

There will be a redundancy, because I could only go as far as one up level to identify a single field in a XML with my previous method, and the 'disambiguator' is three levels above.

Is there a practical way to reach with no mistake a single field given all the above (or at least a few) fields ?